The simplest possible MapReduce job optimization is to perform which of these actions?
Implement a custom Mapper.
Add more master nodes.
Implement optimized InputSplits.
Add more DataNodes.